There is no consensus among health economists about the specification of individual preferences that best fits observed behaviors. One crucial point of disagreement concerns the sign of the interaction between health and wealth in individual utility functions. We propose here a simple, static, model where individuals have only one decision to make concerning their future health ( say to get a flu shot). We derive very simple testable implications that allow to discriminate between three specifications that are commonly used in theoretical models: additive separability between health and consumption, multiplicative separability, and existence of a monetary equivalent of health. We also derive several implications on the pattern of the demand for health insurance.